In the midst of the rapid technological transformation defining the twenty-first century, artificial intelligence and natural language processing have ushered in a profound shift in special education, especially for students with intellectual disabilities. This group is among those most in need of sophisticated technological tools that extend well beyond conventional applications, allowing them to interact with the learning environment in ways that are both flexible and closely attuned to their personal requirements.

Within this context, cognitive computing emerges as a powerful technological paradigm that simulates core human cognitive processes, including linguistic understanding, adaptive learning, and real-time decision-making. Its greatest potential is realized when combined with automatic speech recognition (ASR), multi-modal tactile interaction, and natural language generation (NLG). Together, these technologies create immersive educational settings that communicate with students through multiple sensory channels and deliver immediate, performance-sensitive feedback.

Students classified as having moderate intellectual disabilities, those with IQ levels between 55 and 70, encounter fundamental obstacles in acquiring cognitive skills through traditional methods. These difficulties can be outlined as follows:

1) Sustained attention and focus remain elusive when using conventional materials or participating in regular classroom settings [1].

2) Working memory deficits lead to persistent problems in retrieving previously learned information.

3) The ability to distinguish between similar attributes of objects, such as shapes, sizes, and colors, is markedly impaired.

4) Processing multiple stimuli simultaneously proves challenging, thereby weakening imaginative and overall cognitive capacity.

Employing cognitive computing in the education of these students marks a genuine paradigm shift. NLP systems can analyze both vocal and tactile responses in real time, enabling the provision of immediate, individually tailored instructional support. Recent empirical work has consistently shown that integrating these technologies into special education yields substantial gains in instructional effectiveness [2][3].

The integration of artificial intelligence and natural language processing technologies into the educational process has become an essential requirement rather than a luxury, particularly within the domain of special education. Students with moderate intellectual disabilities constitute one of the groups most in need of advanced technological tools that move beyond traditional multimedia applications toward intelligent interactive systems capable of simulating genuine human interaction through fully integrated auditory, tactile, and visual channels.

These students exhibit clear weaknesses in cognitive processes, largely due to chronic attention deficits that limit their ability to perceive and differentiate the attributes of objects such as shapes, sizes, colors, and directions. While most currently available educational software relies on conventional multimedia components, it falls short of providing genuine adaptive responses that address the unique cognitive needs of each individual learner. Consequently, the present study is defined by the absence of educational software programs grounded in cognitive computing and natural language processing within Arabic-language educational environments designed specifically for students with moderate intellectual disabilities.

What is the effectiveness of an educational software program based on cognitive computing and multi-sensory interaction in enhancing selected cognitive skills among a sample of students with moderate intellectual disabilities?

1) There are statistically significant differences between the scores of the experimental and control groups on cognitive activities (identifying lengths, colors, sizes, and the number of shapes) in the post-test in favor of the experimental group when using the proposed program.

2) There are statistically significant differences between the pre-test and post-test scores of the experimental group in favor of the post-test.

This study aimed to:

1) Identify the challenges associated with teaching students with moderate intellectual disabilities through traditional methods.

2) Develop an educational software program that draws upon cognitive computing and natural language processing, incorporating automatic speech recognition, tactile interaction, and the generation of adaptive responses.

3) Evaluate the effectiveness of the proposed software in improving cognitive skills, specifically color discrimination, size measurement, length measurement, and basic arithmetic operations, among the study sample.

Recent years have witnessed a marked growth in research examining the application of artificial intelligence and natural language processing techniques in special education. The following sections present the most prominent and well-documented recent studies relevant to the current work.

First: Artificial Intelligence and Intelligent Tutoring Systems for Intellectual Disabilities

Hopcan et al. [2] conducted a comprehensive systematic review that tracked the use of intelligent tutoring systems (ITS) in supporting students with intellectual disabilities. The review concluded that these systems contribute meaningfully to monitoring academic performance and delivering personalized instructional interventions, with their effectiveness significantly enhanced when integrated with NLP technologies.

In the same vein, Barua et al. [3], who reviewed 26 studies on intelligent assistive tools for students with neurodevelopmental disorders, demonstrated that adaptive artificial intelligence applications produced clear positive outcomes in improving cognitive skills. They further noted that real-time feedback generated through NLP models can reduce learning time by 20% - 35%.

Alsolami [4] carried out an experimental study involving 70 students aged 9 -12 years with mild intellectual disabilities in Jeddah, Saudi Arabia. The results revealed that artificial intelligence-based interventions led to statistically significant improvements in academic performance across all measured domains, with effect sizes ranging from moderate (η² = 0.685) to large (η² = 0.921). These findings underscore the transformative impact of intelligent technologies in educating this population. Earlier work by Mostafa et al. [5] similarly demonstrated that purpose-built educational software produced measurable gains in cognitive skills among children with moderate intellectual disabilities, providing an important precedent for the present intervention.

Second: Cognitive Computing and Adaptive Learning Technologies

Halkiopoulos and Gkintoni [6] provided evidence of the effectiveness of personalized learning models rooted in cognitive neuroscience for enhancing learning outcomes among students with special educational needs. Their work concluded that combining artificial intelligence with Universal Design for Learning (UDL) principles creates more responsive and inclusive educational environments that better accommodate individual differences.

These findings align closely with the integrative review by Dumitru et al. [7], which synthesized 27 studies on the impact of AI-supported assistive technologies on the academic performance of students with disabilities. The review identified five major themes: personalized learning, benefits of AI-enabled assistive technologies, adoption challenges, institutional barriers, and best practices for implementation.

Third: Assistive Technologies and Tactile-Auditory Interaction

In their systematic review, Fernández-Batanero et al. [1] examined the role of assistive technologies in promoting the inclusion of students with disabilities. They found that AI-driven auditory and tactile tools substantially improve communication abilities and cognitive interaction, particularly when deployed in learning contexts tailored to each student’s unique characteristics.

In a broader systematic review, Voultsiou and Moussiades [8] analyzed 139 studies on the use of artificial intelligence, virtual reality, and large language models in special education. They reported that integrating these technologies strengthens personalized learning, increases social interaction, and accelerates cognitive development among students with special needs. The authors specifically highlighted interactive intelligent characters as a particularly promising avenue in this field. In the Arabic-language context, Elkhalifa and Elhassan [9] demonstrated that auditory-perception-based interfaces substantially improve digital usability for individuals with sensory impairments, reinforcing the value of multi-sensory adaptive design.

Fourth: Theoretical Frameworks for Artificial Intelligence and Intellectual Disability

Almufareh et al. [10] presented a detailed theoretical framework concerning artificial intelligence and intellectual disability. Their analysis focused on the contributions of NLP, speech-to-text conversion, and behavioral bio-analysis in supporting the learning process for individuals with intellectual disabilities. They emphasized that combining immediate auditory feedback with interactive learning environments represents one of the most effective strategies for enhancing cognitive efficiency. The imperative for accessible Arabic-language tools is further underscored by Mustafa et al. [11], who designed a CAPTCHA system specifically adapted for visually impaired Arabic-speaking users, illustrating the broader demand for inclusive Arabic digital interfaces. Furthermore, Alnahdi et al. [12] emphasized the necessity of developing personalized curricula and inclusive practices for students with intellectual disabilities in the Saudi Arabian educational context, further reinforcing the need for adaptive and culturally responsive educational tools of the kind proposed in this study.

In the context of designing the proposed educational software, Figure 1Figure 1 presents the system components. The architecture of the interactive system adopts a comprehensive multi-layered framework that fully embodies the principles of cognitive computing and natural language processing. This structure ensures immediate and adaptive responses that are precisely attuned to the individual cognitive needs of students with moderate intellectual disabilities.

The architecture comprises four interconnected primary layers. The presentation layer features a simple yet engaging visual user interface, complemented by the interactive character known as “the Magic Teacher”, which delivers auditory instructions and provides emotional encouragement. The multi-sensory interaction layer integrates automatic speech recognition (ASR) with natural language processing (NLP) models and natural language generation (NLG), enabling the system to capture both vocal and tactile responses and convert them into commands suitable for real-time processing. The third layer consists of the cognitive

Figure 1. System architecture.

computing and artificial intelligence engine (AI Engine), which analyzes performance patterns in real time and automatically adjusts the level of difficulty, visual stimuli, and the frequency of prompts according to each student’s personal cognitive indicators. Finally, the data and content layer includes an individualized student database for tracking progress, along with a dedicated library of educational content focused on core cognitive skills, color recognition, size and length measurement, and basic arithmetic operations.

In addition, the system includes an adaptive positive reinforcement module that activates celebratory visual and auditory environments upon successful task completion. Data flows smoothly across these layers: the student begins the interaction through either vocal or tactile input; the AI engine then evaluates the response in real time and generates immediate feedback through the interactive character. Subsequently, it proposes the next instructional pathway according to the student’s cumulative performance history. In this way, the learning process shifts from a traditional didactic model to an immersive, dialogic experience that is finely tuned to each learner’s individual rhythm.

This design is not merely a technical blueprint; it represents the outcome of a thoughtful and in-depth consideration of the distinctive characteristics of moderate intellectual disability. By emphasizing the strengthening of sustained attention, working memory, and sensory discrimination through multiple interactive channels, the system also allows the instructor to review detailed performance reports and make manual adjustments to learning objectives when needed. As a result, the platform functions as a truly effective tool for cognitive empowerment in a specially designed Arabic-language educational environment.

To verify the study hypothesis, the scores of both groups on the cognitive skills test were recorded. The test was developed specifically for the present study to assess four domains aligned with the instructional content: color recognition (10 items), size discrimination (10 items), length measurement (10 items), and basic counting operations (10 items), yielding a total of 40 items and a maximum score of 40 points. Each item was scored dichotomously (1 = correct, 0 = incorrect), and the composite score was computed as the sum of correct responses. Content validity was established through a panel of seven specialists in special education and cognitive assessment, who reviewed and approved all items for domain relevance and developmental appropriateness. Reliability was estimated prior to the main study using a test-retest procedure with a comparable group of six students over a two-week interval, yielding a Pearson correlation coefficient of r = 0.87, indicating acceptable stability. Table 2 presents the post-test results for both groups:

Table 2. Post-test results comparing experimental and control groups.

The results clearly indicate that the experimental group achieved a statistically significant superiority. This superiority can be attributed to the following factors:

Effectiveness of the voice-enabled interactive character: Employing the Magic Teacher to deliver instructions and receive responses significantly enhanced students’ attention and concentration, transforming the learning experience into a genuine dialogue rather than passive rote instruction.Enhanced tactile interaction: The system enabled learners to interact physically with abstract cognitive concepts such as length and size. This hands-on engagement deepened conceptual understanding and accelerated information retrieval.Adaptive positive reinforcement: The celebratory reward environment fostered intrinsic motivation and markedly reduced students’ resistance to retrying after making errors.Immediate cognitive adaptation: The system’s capacity to adjust content difficulty in real time allowed each student to learn at a pace that matched their individual cognitive rhythm.

These findings align with recent studies highlighting the effectiveness of intelligent learning systems and natural language processing in promoting cognitive development [2]-[4][6][8].

Within-Group Analysis: Hypothesis B:

To address Hypothesis B, a paired-samples t-test was conducted comparing the experimental group’s pre-test and post-test scores on the cognitive skills measure. The experimental group entered the study with a mean pre-test score of 9.43 (SD = 1.81), reflecting a baseline consistent with the group’s initial equivalence with the control condition. Following the intervention, the same group attained a mean post-test score of 17.88 (SD = 1.49), representing a mean gain of 8.45 points. The paired t-test yielded t (6) = 9.12, p < 0.001, confirming that the improvement from pre-test to post-test was statistically significant. This result provides direct support for Hypothesis B and demonstrates that participation in the AI-based software program was associated with a meaningful within-group cognitive gain over the course of the intervention.

This study targeted students with moderate intellectual disabilities and addressed the challenges of equipping them with basic cognitive skills. The proposed software integrates cognitive computing techniques, natural language processing, and tactile interaction within an immersive environment centered on an interactive character. These features contributed to improving the cognitive abilities of the experimental group and enhancing their self-confidence. In light of these findings, the researchers recommend the following:

1) Expanding the use of voice-enabled interactive characters in special education institutions, with a focus on integrating immediate auditory and tactile feedback.

2) Providing training for teachers and instructors on the use of artificial intelligence-based educational systems and the interpretation of their adaptive outputs.

3) Encouraging developers to design educational platforms for students with intellectual disabilities that rely on Arabic NLP, while taking into account psychological and physiological dimensions at every stage of the design process.

4) Conducting comparative studies of different NLP models (adapted BERT and GPT) to determine the most suitable models for Arabic educational contexts.

5) Expanding research into the impact of AI-driven adaptive positive reinforcement on intrinsic motivation among students with intellectual disabilities.